Fix tessellation fallback recovery

Author: yu-ta-sato

city2graph/morphology.py

@@ -2338,9 +2338,17 @@ def _add_building_info(
         joined["index_right"] = None
 
     if _SOURCE_BUILDING_INDEX_COL in joined.columns:
+        # Fallback cells are matched exactly to their source building, so extra
+        # sjoin matches only duplicate the cell's row and must be dropped. All
+        # operations are positional: the sjoin can duplicate index labels when a
+        # cell contains several building points, which breaks label alignment.
+        source_mask = joined[_SOURCE_BUILDING_INDEX_COL].notna().to_numpy()
+        redundant = source_mask & joined.index.duplicated(keep="first")
+        if redundant.any():
+            joined = joined.loc[~redundant]
+            source_mask = joined[_SOURCE_BUILDING_INDEX_COL].notna().to_numpy()
         source_index = joined[_SOURCE_BUILDING_INDEX_COL]
-        joined["index_right"] = source_index.combine_first(joined["index_right"])
-        source_mask = source_index.notna()
+        joined["index_right"] = np.where(source_mask, source_index, joined["index_right"])
         for column in building_columns:
             joined.loc[source_mask, column] = (
                 source_index.loc[source_mask]

city2graph/utils.py

@@ -4496,7 +4496,18 @@ def _repair_or_drop_degenerate_enclosures(
             "grid_size=1e-3.",
             len(broken),
         )
-        tessellation = run_tessellation(**{**overrides, "grid_size": 1e-3})
+        retry_overrides = {**overrides, "grid_size": 1e-3}
+        try:
+            tessellation = run_tessellation(**retry_overrides)
+        except TypeError as e:
+            if "incorrect geometry type" not in str(e) or "simplify" in kwargs:
+                raise
+            logger.warning(
+                "Tessellation retry failed boundary simplification (%s); retrying with "
+                "simplify=False.",
+                e,
+            )
+            tessellation = run_tessellation(**{**retry_overrides, "simplify": False})
         broken = _overfilled_enclosures(tessellation, enclosures)
     if broken:
         logger.warning(

tests/test_graph.py

@@ -2315,12 +2315,38 @@ def test_osmnx_shaped_multigraph_gdfs_round_trip_deduplicates_symmetrized_edges(
         )
 
     def test_osmnx_downloaded_multidigraph_smoke(self) -> None:
-        """Downloaded OSMnx MultiDiGraph round-trips through PyG with keyed edges."""
-        graph = ox.graph_from_bbox(
-            (-0.128, 51.501, -0.124, 51.503),
-            network_type="drive",
-            simplify=True,
-            retain_all=False,
+        """OSMnx MultiDiGraph round-trips through PyG with keyed edges."""
+        graph = nx.MultiDiGraph()
+        graph.graph["crs"] = "EPSG:4326"
+        graph.add_node(101, x=-0.128, y=51.501)
+        graph.add_node(202, x=-0.126, y=51.502)
+        graph.add_node(303, x=-0.124, y=51.503)
+        graph.add_edge(
+            101,
+            202,
+            key=0,
+            osmid=10,
+            length=100.0,
+            geometry=LineString([(-0.128, 51.501), (-0.126, 51.502)]),
+            oneway=True,
+        )
+        graph.add_edge(
+            202,
+            303,
+            key=0,
+            osmid=11,
+            length=110.0,
+            geometry=LineString([(-0.126, 51.502), (-0.124, 51.503)]),
+            oneway=True,
+        )
+        graph.add_edge(
+            202,
+            101,
+            key=0,
+            osmid=12,
+            length=100.0,
+            geometry=LineString([(-0.126, 51.502), (-0.128, 51.501)]),
+            oneway=True,
         )
         nodes, edges = ox.graph_to_gdfs(graph)
 

tests/test_morphology.py

@@ -431,6 +431,48 @@ def test_include_unenclosed_buildings_is_opt_in(
         assert len(opt_in_nodes["place"]) == 2
         assert "fallback_1" in opt_in_nodes["place"].index
 
+    def test_keep_buildings_with_fallback_and_multi_building_cell(
+        self,
+        sample_crs: str,
+        monkeypatch: pytest.MonkeyPatch,
+    ) -> None:
+        """Fallback cells coexist with cells containing several buildings."""
+        buildings = gpd.GeoDataFrame(
+            {"building_id": ["outer", "nested", "unenclosed"]},
+            geometry=[
+                Polygon([(0, 0), (1, 0), (1, 1), (0, 1)]),
+                Polygon([(0.25, 0.25), (0.75, 0.25), (0.75, 0.75), (0.25, 0.75)]),
+                Polygon([(10, 0), (11, 0), (11, 1), (10, 1)]),
+            ],
+            crs=sample_crs,
+        )
+        segments = gpd.GeoDataFrame(
+            geometry=[LineString([(-1, -1), (2, -1)])],
+            crs=sample_crs,
+        )
+
+        monkeypatch.setattr(
+            "city2graph.morphology.create_tessellation",
+            _enclosed_cell_or_fail_on_fallback,
+        )
+        monkeypatch.setattr(
+            "city2graph.morphology._filter_adjacent_tessellation",
+            _passthrough_adjacent_filter,
+        )
+
+        nodes, _ = morphological_graph(
+            buildings,
+            segments,
+            keep_buildings=True,
+            include_unenclosed_buildings=True,
+        )
+
+        place_ids = list(nodes["place"].index)
+        # The enclosed cell holds both overlapping buildings; the fallback cell
+        # for the unenclosed building is matched exactly once to its source.
+        assert place_ids.count("fallback_2") == 1
+        assert place_ids.count("enclosed") == 2
+
     def test_morphological_graph_passes_tessellation_limit(
         self,
         sample_crs: str,

tests/test_utils.py

@@ -363,6 +363,57 @@ def fake_enclosed_tessellation(**kwargs: object) -> gpd.GeoDataFrame:
         assert result.geometry.area.sum() == pytest.approx(enclosure.area)
         assert "overlapping cells" in caplog.text
 
+    def test_enclosed_tessellation_overlap_retry_falls_back_to_simplify_false(
+        self,
+        sample_buildings_gdf: gpd.GeoDataFrame,
+        sample_segments_gdf: gpd.GeoDataFrame,
+        caplog: pytest.LogCaptureFixture,
+        monkeypatch: pytest.MonkeyPatch,
+    ) -> None:
+        """A geometry-type error during the overlap retry should retry simplify=False."""
+        enclosure = Polygon([(0, 0), (10, 0), (10, 10), (0, 10)])
+        monkeypatch.setattr(
+            momepy,
+            "enclosures",
+            lambda **_kwargs: gpd.GeoDataFrame(
+                {"eID": [0]},
+                geometry=[enclosure],
+                crs=sample_buildings_gdf.crs,
+            ),
+        )
+
+        sane_cells = [
+            Polygon([(0, 0), (5, 0), (5, 10), (0, 10)]),
+            Polygon([(5, 0), (10, 0), (10, 10), (5, 10)]),
+        ]
+
+        def fake_enclosed_tessellation(**kwargs: object) -> gpd.GeoDataFrame:
+            # The first run silently degenerates; the coarser grid_size retry
+            # snaps cells into non-polygonal geometry and blows up in
+            # coverage_simplify; only simplify=False succeeds.
+            if kwargs.get("grid_size") and kwargs.get("simplify") is not False:
+                msg = "One of the Geometry inputs is of incorrect geometry type."
+                raise TypeError(msg)
+            cells = sane_cells if kwargs.get("grid_size") else [enclosure, enclosure]
+            return gpd.GeoDataFrame(
+                {"enclosure_index": [0, 0]},
+                geometry=cells,
+                index=[0, 1],
+                crs=sample_buildings_gdf.crs,
+            )
+
+        monkeypatch.setattr(momepy, "enclosed_tessellation", fake_enclosed_tessellation)
+
+        with caplog.at_level("WARNING"):
+            result = utils.create_tessellation(
+                sample_buildings_gdf,
+                primary_barriers=sample_segments_gdf,
+            )
+
+        assert len(result) == 2
+        assert result.geometry.area.sum() == pytest.approx(enclosure.area)
+        assert "retrying with simplify=False" in caplog.text
+
     def test_enclosed_tessellation_drops_persistently_overlapping_enclosures(
         self,
         sample_buildings_gdf: gpd.GeoDataFrame,